Skip to content
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
259 lines (193 sloc) 7.96 KB
=begin pod :kind("Language") :subkind("Language") :category("fundamental")
=TITLE Statement prefixes
=SUBTITLE Prefixes that alter the behavior of a statement or a set of them
Statement prefixes are written in front of a
statement, and change their meaning, their output, or the moment they are going
to be run. Since they have a specific behavior, they are also sometimes specific
to some statement or group of statements.
=head2 X<C<lazy>|lazy (statement prefix)>
As a statement prefix, C<lazy> acts in front of any statement, including C<for>
loops, saving the execution for when the variable they are assigned to is
actually needed.
=for code
my $incremented = 0;
my $var = lazy for <1 2 3 4> -> $d {
say $incremented; # OUTPUT: «0␤»
say eager $var; # OUTPUT: «(0 1 2 3)␤»
say $incremented; # OUTPUT: «4␤»
The C<$incremented> variable is only incremented, that is, the internal part of
the loop is only run when we eagerly evaluate the variable C<$var> that
contains the lazy loop. Eagerness can be applied on a variable in other ways,
such as calling the C<.eager> method on it.
=for code
my @array = lazy { (^3).map( *² ) };
say @array; # OUTPUT: «[...]»
say @array.eager; # OUTPUT: «[0 1 4]␤»
This prefix can also be used L<in front of
C<gather>|/language/control#gather/take> to make the inner statements behave
lazily; in general, any set of statements that returns a value will be made
lazy using this.
=head2 X<C<eager>|eager (statement prefix)>
The C<eager> statement prefix will eagerly return the result of the statements
behind, throwing away laziness and returning the result.
=for code
my $result := eager gather { for 1..3 { say "Hey"; take $_² } };
say $result[0]; # OUTPUT: «Hey␤Hey␤Hey␤1␤»
C<gather> is L<implicitly lazy when bound to a scalar|/syntax/gather%20take>.
However, with C<eager> as a statement prefix it will run all three iterations in
the loop, as shown by the printed "Hey", even if we are just requesting the
first one in a row.
=head2 X<C<hyper>|hyper (statement prefix)>, X<C<race>|race (statement prefix)>
A C<for> loop will automatically serialize any L<C<HyperSeq>|/type/HyperSeq> or
L<C<RaceSeq>|/type/RaceSeq> used in it; on the other hand C<hyper> and C<race>
use (maybe simultaneous) threads to run different iterations in a loop:
=for code
my @a = hyper for ^100_000 { .is-prime }
This code is around 3x faster than the bare C<for>. But there are a couple of
caveats here:
=item The operation inside the loop should take enough time for threading
to make sense.
=item There should be no read or write access to the same data structure inside
the loop. Let the loop produce a result, and assign it.
=item If there's an I/O operation inside the loop, there might be some contention
so please avoid it.
Main difference between C<hyper> and C<race> is the ordering of results. Use
C<hyper> if you need the loop results to be produced in order, C<race> if you
don't care.
=head2 X<C<quietly>|quietly (statement prefix)>
As a statement prefix, C<quietly> suppresses all warnings produced by the
statement it precedes.
=for code
sub marine() {};
quietly say ~&marine; # OUTPUT: «marine␤»
Calling L<C<.Str> on C<code> produces a warning|/type/Code#method_Str>.
Preceding the statement with C<quietly > will just produce the output, the name
of the routine.
=head2 X<C<try>|try (statement prefix)>
If you use C<try> in front of a statement, it will contain the exception
produced in it and store it in the C<$!> variable, just like when L<it's used in
front of a block|/language/exceptions#try_blocks>.
=for code
try [].pop;
say $!; # OUTPUT: «Cannot pop from an empty Array␤..»
=head2 X<C<do>|do (statement prefix)>
C<do> can be used as an statement prefix to disambiguate the statement they
precede; this is needed, for instance, if you want to assign the result of a
C<for> statement. A bare C<for> will fail, but this will work:
=for code
my $counter = 0;
my $result = do for ^5 { $counter++ };
say $counter; # OUTPUT: «5␤»
say $result; # OUTPUT: «(0 1 2 3 4)␤»
C<do> is equivalent, as in other cases, to surrounding a statement with a
parenthesis. It can be used as an alternative with a (possibly more)
straightforward syntax.
=head2 X<C<sink>|sink (statement prefix)>
As in the L<case of the routine|/routine/sink>, C<sink> will run the statement,
throwing away the result. Use it in case you want to run some statement for the
side effects it produces.
=for code
my $counter = 0;
my $result = sink for ^5 { $counter++ };
say $counter; # OUTPUT: «5␤»
say $result; # OUTPUT: «(Any)␤»
The C<sink> statement prefix will also convert C<Failure>s into exceptions:
=for code
sub find-the-number ( Int $n where $n < 10 ) {
if $n == 7 {
return True;
} else {
fail "Not that number" ;
for 1..^10 {
try {
sink find-the-number($_);
say "Found $_" unless $!;
In this case, we will know that the number has been found only when the
C<try> block is not catching an exception.
=head2 X<C<once>|once (statement prefix)>
Within a loop, runs the prefixed statement only once.
=for code
my $counter;
my $result = do for ^5 { once $counter = 0; $counter++ };
say $result; # OUTPUT: «(0 1 2 3 4)␤»
=head2 X<C<gather>|gather (statement prefix)>
C<gather> can be used in front of a statement, receiving and gathering in a list
all data structures emitted from a C<take> run anywhere from that statement:
=begin code
proto sub fact( Int ) {*}
multi sub fact( 1 --> 1 ) {}
multi sub fact( $x ) { take $x * fact( $x-1 ) }
my @factors = gather say fact(13); # OUTPUT: «6227020800»
say @factors;
# OUTPUT: «[2 6 24 120 720 5040 40320 362880 3628800 ...]»
=end code
In this example, C<gather> precedes C<say>, which prints the first result of the
factorial; at the same time, it's harvesting the result from every call to
C<fact>, which goes to C<@factor>.
=head2 X<C<start>|start (statement prefix)>
As a statement prefix, C<start> behaves in the same way as L<in front of a
block|/language/control#flow%29_start>, that is, it runs the statement
asynchronously, and returns a promise.
=begin code
proto sub fact( Int ) {*}
multi sub fact( 1 --> 1 ) {}
multi sub fact( $x ) { $x * fact( $x-1 ) }
my @promises = gather {
for <3 4> {
take start fact( 10 ** $_ );
say await @promises;
=end code
The L<C<Promise>s|/type/Promise> created by start are gathered in an array,
which returns the result of the operation once the promises have been fulfilled.
=head2 X<C<react>|react (statement prefix)>
C<react> can be used in concurrent programs to create blocks of code that run
whenever some event occurs. It L<works with blocks|/syntax/react>, and also as a
statement prefix.
=begin code
my Channel $KXGA .= new;
for ^100 {
$KXGA.send( (100000..200000).pick );
my @sums = ( start react whenever $KXGA -> $number {
say "In thread ", $*;
say "→ ", (^$number).sum;
} ) for ^10;
start { sleep 10; $KXGA.close(); }
await @sums;
=end code
In this case C<react> prefixes C<whenever>, which makes a long sum with every
number read from a channel.
=head2 X<C<supply>|supply (statement prefix)>
The keyword C<supply> creates
L<on-demand supplies|/language/concurrency#index-entry-supply_(on-demand)>
that you can tap. It pairs with C<emit>, which can be used anywhere from within
a C<supply> prefixed statement.
=begin code
my &cards = -> {
my @cards = 1..10 X~ <♠ ♥ ♦ ♣>;
emit($_) for @cards.pick(@cards.elems);
my $supply = supply cards;
$supply.tap( -> $v { say "Drawing: $v" });
$supply.tap( -> $v { say "Drawing: $v" }, done => { say "No more cards" });
# [...]
# Drawing: 1♥
# Drawing: 7♥
# Drawing: 9♥
# No more cards
=end code
In this example, C<supply> acts as prefix of the previously defined C<cards>
routine. It would very well be defined as a block, but giving it a name in this
case might increase legibility or simply give the responsibility of defining it
to other module.
=end pod
# vim: expandtab softtabstop=4 shiftwidth=4 ft=perl6
You can’t perform that action at this time.